First early Flight Video only for test that all work fine ... a windy day and need to setup PID because with more power the old one is not good:

Dear Friends,
i'm happy to inform you that our new product Multipilot32F4 and NAVY 2012 is ready
for beta testing for a special price of 199 euro.
We presented it at the beginning of year .We finish to port low
level library from STM32F1 to STM32F4: We rewrite old libmaple for STM32F1 by scratch porting
all lib to STM32F4. (thank's Alberto V for his work on library)
So the AP_LIB have same interface of standard AP_LIB for arduino and
the Arducopter32 is very similar to original revision for Arduino
platform , share 98% with original code.

Original Blog Post : http://www.virtualrobotix.com/profiles/blogs/vr-multipilot32f4-the-most-powerful-auto-pilot-in-the-world-compa

The firmware that i using in my fly test is Arducopter 2.5.4 so it's last available.

I'm doing my first fly test in these day with success . A video will be availabe in the next days

The characteristics of Multipilot32F4 are:

Hardware:

ARM Cortex-M4 processor STM32F407VET6. 168 Mhz
Flash 1024 Kbytes RAM 196 Kbytes
16-bit Timer 4
SPI 2 (ADC Interface , MicroSD connection Option)
I ² C 2 (First I2C (sensor), Second I2C control until ESC 12)
USART 5 (GPS, DEBUG Console, XBee Pro Telemetry)
USB 1 (Upload Firmware, Debug Console, Power Board for Debug)
CAN 1 (Interconnection with Professional ESC 1 Mbit update rate)
8 PWM Output Bit 16 (ESC / Servo Control)
7 PWM Input 16 Bit (RC Input Channel, accept PPM SUM)
8 Analog Input 12 Bit.
Professional 4 layers PCB.
DC: DC 30 V (6s Lipo): 5 volts and 3.3 volts
MPU6000 Imu Sensor 3 gyro and 3 acc on board

The NAVY 2012

HMC5883 3 axis magnetic sensor
EEPROM for data storage
MS6511 high precision barometer
MTEK GPS with custom firmware for arducopter.
Dataflash on board.
SDIO SD Card
SPI SD Card that support FAT file systems. (Optional)

More info about the boards are available here :

http://www.virtualrobotix.com/page/multipilot32f4-v3-0
The board is yet in production and available the next stock could be
available in the middle of may.

There are a lot of tool for development :

A very light toolchain for peolpe that like command line ... a lot in hard developer community.
A VRIDE Pro that is an eclipse with support for JTAG debugging .. we using ST-LINK.
An VRIDE Std that will be available in some week for end user .

Is possible to use the DFU on usb for upgrade firmware without problem.
This is the link for download last revision of enviroment :http://code.google.com/p/multipilot32/downloads/detail?name=VR%20MP32F4%20%20IDE%20LIB%20FIRMWARE%2001-5-2012%202.6%20%20beta.rar&can=2&q=#makechanges

This is the link of the thread on virtualrobotix forum:
http://www.virtualrobotix.com/forum/topics/multipilot32-v3-f1-f4-and-new-vr-ide-pro

In our test we found more of 12x respect to standard Arduino Mega hardware platform.

The spare time cicle in 5 second is 3610 for APM and 42321 for an Multipilo32F4.

In this test is not yet available the FPU acceleration only by different clock 16 mhz for arduino , 168 for Multipilot32F4

In this video you can see the different whe use FPU ON or FPU OFF math acceleragion going up 17x


This is the new benchmark
Arduino APM1 e APM2
PerfMon:            cpu%          mils #called Hz
loop                     33.45%  96 3614 0hz
read_AHR             34.09%    98       29 0hz
FiftyHZ_Loop         2.88%         8        15 0hz
Update_GPS           0.56%                  1       15 0hz
setupA                   0.00%              0   0 0hz
unexplained         29.03% 83

STM32F1                        MP32F1
PerfMon:                       cpu%        mils #called    Hz
loop                            35.81%  112     18357   0hz
read_AHR               23.53%  73      130             0hz Fif y
FiftyHZ_Loop   H           7.54   54      15              0hz
SuperFast_Loop         3.04%   9       392             0hz
Update_GPS             1.10%   3       15              0hz
ReadBarometer          0.07%   0       3               0hz
setupA                 0.00%   0       0               0hz
unexplained             18.91%  59

STM32F4                 MP32F4 ( NOW I2C IS OK )
PerfMon:                 cpu%    mils    #called Hz
loop                       46.88%             131 42321 0hz
read_AHR               11.01%  30      138     0hz
SuperFast_Loop         8.45%   23      519     0hz
FiftyHZ_Loop           3.00%   8       15      0hz
Update_GPS             0.13%   0       15      0hz
ReadBarometer          0.04%   0       3       0hz
setupA                 0.00%   0       0       0hz
unexplained            30.49%  85

So if someone would join the development is welcome :)

Best

Roberto Navoni

Hi Adam,

so i think that the MP32F4 is a good platform for developing and testing a lot of different kind of fly algorithm.

I develop it because during my last experience with MP32F1 and MP8 first. I join some university project for develop prototype for different kind of application.

So the MP32F4 is a good platform for fast prototyping but is good also for develop final product.

The advantages is that you have a power of Arm Cortex M4 with the semplicity of arduino compatible framework.

So at the moment i'm investigate to identify developer that would check ekf on MP32F4 . Join our dev team davide that is in R&D company of ST Micro Electronics and they would help us in our development.

They developing robot platform for demostrate ST component. So the next week will be available a new sensor for us with the support of ST 3 gyro and 3 acc as MPU6000 . We check it and going in fly :)

Could be possible that they support us for EKF on MP32F4 . This is only a first step .

My idea is to continue to improve our great opensource lib AP_XXX with this new feature that could be have some limits if i use an arduino processor , instead with MP32F4 you don't have any kind of limit for this kind of improvment.

Then MP32F4 don't have some limits that had MP32F1 for example STM32F103 don't support the 2 bit stop odd parity protocol that is important for use directly the telemetry of some device. On F4 we have I2c or CAN bus for develop interprocessing comunication protocol for connect 3 cpu togheter and try to develop a failsafe system for example. On F4 we have a usb host or LAN so could be possible implement an efficient system for interconnect ROS compatible board for advanced application.

In my mind the actual framework of lib could be improve with lan , usb host and a lot of advanced filter , attitude and so' on algorithm and feature.

Another interesting feature of F407 is the video input. Is possible to implement a lot of feature on that micro with a cam ... could be possible reimplement the optical loiter with more power and functionality ... data fusion directly with attitute sensor or use two syncronized board for stereo vision .... 3d attitude using a stereo vision in realtime could be interesting feature. :)

This is what i have in my mind and the reason because i decide to work on this kind of component .

Best

Roberto

You need to be a member of diydrones to add comments!

Join diydrones

Comments

Guillaume Helle May 16, 2012 at 5:35am

On F4 we have a usb host or LAN so could be possible implement an efficient system for interconnect ROS compatible board for advanced application

Does it mean we can connect an Android phone, an USB webcam or other peripherals?
Moderator

Roberto Navoni May 2, 2012 at 1:39pm

Hi John ,

the frame is one of us experiment a 3d printed frame design by a Fox Team User check on www.virtualrobotix.com for more info . It's first experiment . The result it's good . It's very flexible and resist to small crash .. i lost only small landing gear in four month of test and development of MP32F4 and MP32F1.

Doing test on this revision of frame i have some improvment for next release . The next could be water proof the main problem is battery and the temperature of esc .

Best

Roberto
Moderator

Roberto Navoni May 2, 2012 at 1:36pm

Hi Robert,

if you check this channel you can see the possible development using STM32F1-F4 with simple algorithm of harris , blob detection , face detection for simple robotic application : https://www.youtube.com/user/StAstRobotics/videos

Is possible to develop a shield on MP32F4 with a large DRAM or high Speed Static RAM end use DMA for fast data transfer from sensor to ram .. then inside static ram is possible to use high speed algorithm for blob detection , harris (edge detection) vector speed .. So i ask to Davide if is possible develop a 3d attitude sensor using stereo camera .

Best

Roberto
Francisc Bereczky May 2, 2012 at 1:10pm

The clock frequency being 168 MHZ with an efficient software I think is possible. Maybe 196 K of RAM is not enough.
Rob_Lefebvre May 2, 2012 at 1:03pm

Is the chip capable of doing onboard video processing, like object tracking? Just curious.
Jack Crossfire May 2, 2012 at 12:11pm

Wish it did have 192kb of SRAM. The STM32F4 is a great chip. It can JPEG compress 320x240 video at 30fps.
John Moore May 2, 2012 at 12:00pm

Is that a waterproof frame?
Moderator

Roberto Navoni May 2, 2012 at 9:16am

Hi Adam,

so i think that the MP32F4 is a good platform for developing and testing a lot of different kind of fly algorithm.

I develop it because during my last experience with MP32F1 and MP8 first. I join some university project for develop prototype for different kind of application.

So the MP32F4 is a good platform for fast prototyping but is good also for develop final product.

The advantages is that you have a power of Arm Cortex M4 with the semplicity of arduino compatible framework.

So at the moment i'm investigate to identify developer that would check ekf on MP32F4 . Join our dev team davide that is in R&D company of ST Micro Electronics and they would help us in our development.

They developing robot platform for demostrate ST component. So the next week will be available a new sensor for us with the support of ST 3 gyro and 3 acc as MPU6000 . We check it and going in fly :)

Could be possible that they support us for EKF on MP32F4 . This is only a first step .

My idea is to continue to improve our great opensource lib AP_XXX with this new feature that could be have some limits if i use an arduino processor , instead with MP32F4 you don't have any kind of limit for this kind of improvment.

Then MP32F4 don't have some limits that had MP32F1 for example STM32F103 don't support the 2 bit stop odd parity protocol that is important for use directly the telemetry of some device. On F4 we have I2c or CAN bus for develop interprocessing comunication protocol for connect 3 cpu togheter and try to develop a failsafe system for example. On F4 we have a usb host or LAN so could be possible implement an efficient system for interconnect ROS compatible board for advanced application.

In my mind the actual framework of lib could be improve with lan , usb host and a lot of advanced filter , attitude and so' on algorithm and feature.

Another interesting feature of F407 is the video input. Is possible to implement a lot of feature on that micro with a cam ... could be possible reimplement the optical loiter with more power and functionality ... data fusion directly with attitute sensor or use two syncronized board for stereo vision .... 3d attitude using a stereo vision in realtime could be interesting feature. :)

This is what i have in my mind and the reason because i decide to work on this kind of component .

Best

Roberto
Moderator

Roberto Navoni May 2, 2012 at 9:00am

Hi Monroe,

For the developer that want join us on MP32F4 development contact virtualrobotix@gmail.com for request more info. Or join dev group inside www.virtualrobotix.com and send a pm to me.
Developer

Adam Rivera May 2, 2012 at 8:41am

How do you intend on leveraging the additional hardware capabilities?

This reply was deleted.

Blogs

VR Multipilot32F4, (Arm Cortex M4) the most powerful Auto Pilot in the world with STM32F4 compatible with arduino. It's ready to Fly.

The characteristics of Multipilot32F4 are:

Hardware:

Comments